CarND Project 4 - Advanced Lane Lines Detection

Initialization

In [3]:
import cv2
import glob
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
import pickle

%matplotlib inline

1. Camera Calibration

In [2]:
%matplotlib qt

# prepare object points, like (0,0,0), (1,0,0), (2,0,0) ....,(6,5,0)
objp = np.zeros((6*9,3), np.float32)
objp[:,:2] = np.mgrid[0:9,0:6].T.reshape(-1,2)

# Arrays to store object points and image points from all the images.
objpoints = [] # 3d points in real world space
imgpoints = [] # 2d points in image plane.

# Make a list of calibration images
images = glob.glob('./camera_cal/calibration*.jpg')

# Step through the list and search for chessboard corners
for fname in images:
    img = cv2.imread(fname)
    img_size = (img.shape[1], img.shape[0])
    gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)

    # Find the chessboard corners
    ret, corners = cv2.findChessboardCorners(gray, (9,6), None)

    # If found, add object points, image points
    if ret == True:
        objpoints.append(objp)
        imgpoints.append(corners)

        # Draw and display the corners
        img = cv2.drawChessboardCorners(img, (9,6), corners, ret)
        cv2.imshow('img', img)
#         cv2.waitKey(500)

cv2.destroyAllWindows()

# Do camera calibration given object points and image points
ret, mtx, dist, rvecs, tvecs = cv2.calibrateCamera(objpoints, imgpoints, img_size,None,None)

# Save the camera calibration result for later use (we won't worry about rvecs / tvecs)
dist_pickle = {}
dist_pickle["mtx"] = mtx
dist_pickle["dist"] = dist
pickle.dump(dist_pickle, open("camera_cal/camera_cal_pickle.p", "wb"))

Apply a distortion correction to raw images.

In [3]:
%matplotlib inline

# Test undistortion on an image
img = cv2.imread('camera_cal/calibration1.jpg')
img_size = (img.shape[1], img.shape[0])

dst = cv2.undistort(img, mtx, dist, None, mtx)
cv2.imwrite('camera_cal/test_undist.jpg',dst)

# Visualize undistortion
f, (ax1, ax2) = plt.subplots(1, 2, figsize=(20,10))
ax1.imshow(img)
ax1.set_title('Original Image', fontsize=30)
ax2.imshow(dst)
ax2.set_title('Undistorted Image', fontsize=30)
Out[3]:
<matplotlib.text.Text at 0x11d740160>
In [4]:
%matplotlib inline

with (open("camera_cal/camera_cal_pickle.p", "rb")) as openfile:
    try:
        dist_pickle = pickle.load(openfile)
        mtx = dist_pickle["mtx"]
        dist = dist_pickle["dist"]
    except EOFError:
        print(EOFError)
            
src_img = cv2.imread('test_images/test1.jpg')

dst_img = cv2.undistort(src_img, mtx, dist, None, mtx)

src_img = cv2.cvtColor(src_img, cv2.COLOR_BGR2RGB)
dst_img = cv2.cvtColor(dst_img, cv2.COLOR_BGR2RGB)

# Visualize undistortion
f, (ax1, ax2) = plt.subplots(1, 2, figsize=(20,10))
ax1.imshow(src_img)
ax1.set_title('Original Image', fontsize=30)
ax2.imshow(dst_img)
ax2.set_title('Undistorted Image', fontsize=30)
Out[4]:
<matplotlib.text.Text at 0x11cb8c198>

2. Image Thresholding

2.1 Gradient Methods

In [5]:
def abs_sobel_thresh(gray, orient='x', sobel_kernel=3, thresh=(0, 255)):
    """Calculate directional gradient & Apply threshold"""
    # Apply x or y gradient with the OpenCV Sobel() function
    # and take the absolute value
    if orient == 'x':
        abs_sobel = np.absolute(cv2.Sobel(gray, cv2.CV_64F, 1, 0))
    if orient == 'y':
        abs_sobel = np.absolute(cv2.Sobel(gray, cv2.CV_64F, 0, 1))
    # Rescale back to 8 bit integer
    scaled_sobel = np.uint8(255*abs_sobel/np.max(abs_sobel))
    # Create a copy and apply the threshold
    binary_output = np.zeros_like(scaled_sobel)
    # Here I'm using inclusive (>=, <=) thresholds, but exclusive is ok too
    binary_output[(scaled_sobel >= thresh[0]) & (scaled_sobel <= thresh[1])] = 1

    # Return the result
    return binary_output

def mag_thresh(gray, sobel_kernel=3, mag_thresh=(0, 255)):
    """Calculate gradient magnitude & Apply threshold"""    
    # Take both Sobel x and y gradients
    sobelx = cv2.Sobel(gray, cv2.CV_64F, 1, 0, ksize=sobel_kernel)
    sobely = cv2.Sobel(gray, cv2.CV_64F, 0, 1, ksize=sobel_kernel)
    # Calculate the gradient magnitude
    gradmag = np.sqrt(sobelx**2 + sobely**2)
    # Rescale to 8 bit
    scale_factor = np.max(gradmag)/255 
    gradmag = (gradmag/scale_factor).astype(np.uint8) 
    # Create a binary image of ones where threshold is met, zeros otherwise
    binary_output = np.zeros_like(gradmag)
    binary_output[(gradmag >= mag_thresh[0]) & (gradmag <= mag_thresh[1])] = 1

    # Return the binary image
    return binary_output

def dir_threshold(gray, sobel_kernel=3, thresh=(np.pi/4.5, np.pi/2.5)):
    """Calculate gradient direction & Apply threshold"""    
    # Calculate the x and y gradients
    sobelx = cv2.Sobel(gray, cv2.CV_64F, 1, 0, ksize=sobel_kernel)
    sobely = cv2.Sobel(gray, cv2.CV_64F, 0, 1, ksize=sobel_kernel)
    # Take the absolute value of the gradient direction, 
    # apply a threshold, and create a binary image result
    absgraddir = np.arctan2(np.absolute(sobely), np.absolute(sobelx))
    binary_output =  np.zeros_like(absgraddir)
    binary_output[(absgraddir >= thresh[0]) & (absgraddir <= thresh[1])] = 1

    # Return the binary image
    return binary_output

Testing Gradient Methods Only

In [6]:
# Choose a Sobel kernel size
ksize = 7 # Choose a larger odd number to smooth gradient measurements

x_thresh_min = 50
x_thresh_max = 150

y_thresh_min = 50
y_thresh_max = 150

mag_thresh_min = 50
mag_thresh_max = 150

# dir_thresh = (np.pi/6, np.pi/2)
# dir_thresh = (np.pi/4.5, np.pi/3)
dir_thresh = (np.pi/4.5, np.pi/2.5)

x_thresh = (x_thresh_min, x_thresh_max)
y_thresh = (y_thresh_min, y_thresh_max)
mag_thresh_val = (mag_thresh_min, mag_thresh_max)

# Apply each of the thresholding functions
# src_img = cv2.imread('test_images/straight_lines1.jpg')
src_img = cv2.imread('test_images/test3.jpg')

src_img = cv2.cvtColor(src_img, cv2.COLOR_BGR2RGB)
hls = cv2.cvtColor(src_img, cv2.COLOR_RGB2HLS).astype(np.float)
l_channel = hls[:,:,1]

gradx = abs_sobel_thresh(l_channel, orient='x', sobel_kernel=ksize, thresh=x_thresh)
grady = abs_sobel_thresh(l_channel, orient='y', sobel_kernel=ksize, thresh=y_thresh)
mag_binary = mag_thresh(l_channel, sobel_kernel=ksize, mag_thresh=mag_thresh_val)
dir_binary = dir_threshold(l_channel, sobel_kernel=ksize, thresh=dir_thresh)

dst_x = np.zeros_like(dir_binary)
dst_y = np.zeros_like(dir_binary)
dst_mag = np.zeros_like(dir_binary)
dst_dir = np.zeros_like(dir_binary)

dst_x[gradx==1]=1
dst_y[grady==1]=1
dst_mag[mag_binary==1]=1
dst_dir[dir_binary==1]=1

combined = np.zeros_like(dir_binary)
combined[((gradx == 1) & (grady == 1)) | ((mag_binary == 1) & (dir_binary == 1))] = 1

f, axs = plt.subplots(3, 2, figsize=(20,15))

axs[0][0].imshow(dst_x, cmap='gray')
axs[0][0].set_title('Grad X Image', fontsize=30)

axs[0][1].imshow(dst_y, cmap='gray')
axs[0][1].set_title('Grad Y Image', fontsize=30)

axs[1][0].imshow(dst_mag, cmap='gray')
axs[1][0].set_title('Grad Mag Image', fontsize=30)

axs[1][1].imshow(dst_dir, cmap='gray')
axs[1][1].set_title('Grad Direction Image', fontsize=30)

axs[2][0].imshow(src_img)
axs[2][0].set_title('Original Image', fontsize=30)

axs[2][1].imshow(combined, cmap='gray')
axs[2][1].set_title('Binary Image', fontsize=30)
Out[6]:
<matplotlib.text.Text at 0x11dbfdef0>

2.2 Adding Color Thresholding

In [6]:
def color_grad_thr(img, ksize=7, s_thresh=(190, 255), sx_thresh=(50, 150), 
#              dir_thresh=(np.pi/4.5, np.pi/2.4)):
             dir_thresh=(np.pi/4.5, np.pi/2.5)):
    img = np.copy(img)
    # Convert to HLS color space and separate the H channel
    hls = cv2.cvtColor(img, cv2.COLOR_RGB2HLS).astype(np.float)
    luv = cv2.cvtColor(img, cv2.COLOR_RGB2LUV).astype(np.float)
#     lab = cv2.cvtColor(img, cv2.COLOR_RGB2LUV).astype(np.float)
    l_channel = hls[:,:,1]
    s_channel = hls[:,:,2]
#     s_channel = luv[:,:,0]
#     s_channel = lab[:,:,2]

    gradx = abs_sobel_thresh(l_channel, orient='x', sobel_kernel=ksize, thresh=sx_thresh)
    grady = abs_sobel_thresh(l_channel, orient='y', sobel_kernel=ksize, thresh=sx_thresh)
    mag_binary = mag_thresh(l_channel, sobel_kernel=ksize, mag_thresh=sx_thresh)
    dir_binary = dir_threshold(l_channel, sobel_kernel=ksize, thresh=dir_thresh)

    dst_x = np.zeros_like(dir_binary)
    dst_y = np.zeros_like(dir_binary)
    dst_mag = np.zeros_like(dir_binary)
    dst_dir = np.zeros_like(dir_binary)

    dst_x[gradx==1]=1
    dst_y[grady==1]=1
    dst_mag[mag_binary==1]=1
    dst_dir[dir_binary==1]=1

    sxbinary = np.zeros_like(gradx)
    sxbinary[(gradx == 1) & (grady == 1)] = 1
    smbinary = np.zeros_like(gradx)
    smbinary[(mag_binary == 1) & (dir_binary == 1)] = 1

    # Threshold color channel
    s_binary = np.zeros_like(s_channel)
    s_binary[(s_channel >= s_thresh[0]) & (s_channel <= s_thresh[1])] = 1
    # Stack each channel
    color_binary = np.dstack((smbinary, sxbinary, s_binary))
    
    combined_binary = np.zeros_like(sxbinary)
    combined_binary[(s_binary == 1) | (sxbinary == 1) | (smbinary == 1)] = 1

    return color_binary, combined_binary

Test color_grad_thr

In [272]:
image = mpimg.imread('test_images/test6.jpg')
color_binary, combined_binary = color_grad_thr(image, s_thresh=(190, 255), sx_thresh=(50, 150))

# Plot the result
f, (ax1, ax2) = plt.subplots(1, 2, figsize=(24, 9))
f.tight_layout()

ax1.imshow(image)
ax1.set_title('Original Image', fontsize=40)

ax2.imshow(color_binary)
ax2.set_title('Pipeline Result', fontsize=40)
plt.subplots_adjust(left=0., right=1, top=0.9, bottom=0.)
3. Perspective Transformation: to rectify binary image ("birds-eye view")
In [12]:
def warper(img, src, dst):
    """Compute and apply perpective transform"""
    img_size = (img.shape[1], img.shape[0])
    M = cv2.getPerspectiveTransform(src, dst)
    warped = cv2.warpPerspective(img, M, img_size, flags=cv2.INTER_NEAREST)  # keep same size as input image

    return warped

src_img = cv2.imread('test_images/straight_lines1.jpg')
# src_img = cv2.imread('test_images/test1.jpg')
img_size = (src_img.shape[1], src_img.shape[0])

src_pts = np.float32(
    [[(img_size[0] / 2) - 62, img_size[1] / 2 + 100],
    [((img_size[0] / 6) - 18), img_size[1]],
    [(img_size[0] * 5 / 6) + 30, img_size[1]],
    [(img_size[0] / 2 + 60), img_size[1] / 2 + 100]])
dst_pts = np.float32(
    [[(img_size[0] / 4), 0],
    [(img_size[0] / 4), img_size[1]],
    [(img_size[0] * 3 / 4), img_size[1]],
    [(img_size[0] * 3 / 4), 0]])

src_img = cv2.undistort(src_img, mtx, dist, None, mtx)

dst_img = warper(src_img, src_pts, dst_pts)

src_img = cv2.cvtColor(src_img, cv2.COLOR_BGR2RGB)
dst_img = cv2.cvtColor(dst_img, cv2.COLOR_BGR2RGB)

# Visualize perpective transform
f, (ax1, ax2) = plt.subplots(1, 2, figsize=(20,10))
cv2.polylines(src_img, np.int32([src_pts]), isClosed=True, color=(255,0,0), thickness=4)#[, lineType[, shift]]]) 
ax1.imshow(src_img)
ax1.set_title('Original Image', fontsize=30)

# dst_pts = dst_pts.reshape(-1,1,2)
cv2.polylines(dst_img, np.int32([dst_pts]), isClosed=True, color=(255,0,0), thickness=4)#[, lineType[, shift]]]) 
ax2.imshow(dst_img)
ax2.set_title('Undistorted Image', fontsize=30)
Out[12]:
<matplotlib.text.Text at 0x11b390c88>

3.1 Testing Color and Gradiant Thresholding and Perspective Transformation on a Curved Lane

In [13]:
src_img = cv2.imread('test_images/test6.jpg')
src_img = cv2.undistort(src_img, mtx, dist, None, mtx)
src_img = cv2.cvtColor(src_img, cv2.COLOR_BGR2RGB)
color_binary, combined_binary = color_grad_thr(src_img, ksize=7, s_thresh=(170, 235), 
                                               sx_thresh=(30, 150))
src_img_wraped = warper(src_img, src_pts, dst_pts)
binary_warped = warper(combined_binary, src_pts, dst_pts)

# Plot the result
f, (ax1, ax2) = plt.subplots(1, 2, figsize=(24, 9))
f.tight_layout()

ax1.imshow(src_img_wraped)
ax1.set_title('Original Wraped', fontsize=40)

ax2.imshow(binary_warped, cmap='gray')
ax2.set_title('color_grad_thr and Wraped Result', fontsize=40)
plt.subplots_adjust(left=0., right=1, top=0.9, bottom=0.)

4. Lane Detection

4.1 Plain Sliding Window Method

In [16]:
# Take a histogram of the bottom half of the image
def lane_detection(binary_warped):
    histogram = np.sum(binary_warped[binary_warped.shape[0]/2:,:], axis=0)
    # Create an output image to draw on and  visualize the result
    out_img = np.dstack((binary_warped, binary_warped, binary_warped))*255

    # Find the peak of the left and right halves of the histogram
    # These will be the starting point for the left and right lines
    side_margin = 50
    midpoint = np.int(histogram.shape[0]/2)
    leftx_base = np.argmax(histogram[side_margin:midpoint])
    rightx_base = np.argmax(histogram[midpoint:-side_margin]) + midpoint

    # Choose the number of sliding windows
    nwindows = 9
    # Set height of windows
    window_height = np.int(binary_warped.shape[0]/nwindows)
    # Identify the x and y positions of all nonzero pixels in the image
    nonzero = binary_warped.nonzero()
    nonzeroy = np.array(nonzero[0])
    nonzerox = np.array(nonzero[1])

    # Current positions to be updated for each window
    leftx_current = leftx_base
    rightx_current = rightx_base
    # Set the width of the windows +/- margin
    margin = 100
    # Set minimum number of pixels found to recenter window
    minpix = 50
    # Create empty lists to receive left and right lane pixel indices
    left_lane_inds = []
    right_lane_inds = []

    # Step through the windows one by one
    for window in range(nwindows):
        # Identify window boundaries in x and y (and right and left)
        win_y_low = binary_warped.shape[0] - (window+1)*window_height
        win_y_high = binary_warped.shape[0] - window*window_height
        win_xleft_low = leftx_current - margin
        win_xleft_high = leftx_current + margin
        win_xright_low = rightx_current - margin
        win_xright_high = rightx_current + margin
        # Draw the windows on the visualization image
        cv2.rectangle(out_img,(win_xleft_low,win_y_low),(win_xleft_high,win_y_high),(0,255,0), 2) 
        cv2.rectangle(out_img,(win_xright_low,win_y_low),(win_xright_high,win_y_high),(0,255,0), 2) 
        # Identify the nonzero pixels in x and y within the window
        good_left_inds = ((nonzeroy >= win_y_low) & (nonzeroy < win_y_high) & (nonzerox >= win_xleft_low) & (nonzerox < win_xleft_high)).nonzero()[0]
        good_right_inds = ((nonzeroy >= win_y_low) & (nonzeroy < win_y_high) & (nonzerox >= win_xright_low) & (nonzerox < win_xright_high)).nonzero()[0]
        # Append these indices to the lists
        left_lane_inds.append(good_left_inds)
        right_lane_inds.append(good_right_inds)
        # If you found > minpix pixels, recenter next window on their mean position
        if len(good_left_inds) > minpix:
            leftx_current = np.int(np.mean(nonzerox[good_left_inds]))
        if len(good_right_inds) > minpix:        
            rightx_current = np.int(np.mean(nonzerox[good_right_inds]))

    # Concatenate the arrays of indices
    left_lane_inds = np.concatenate(left_lane_inds)
    right_lane_inds = np.concatenate(right_lane_inds)

    # Extract left and right line pixel positions
    leftx = nonzerox[left_lane_inds]
    lefty = nonzeroy[left_lane_inds] 
    rightx = nonzerox[right_lane_inds]
    righty = nonzeroy[right_lane_inds] 

    # Fit a second order polynomial to each
    left_fit = np.polyfit(lefty, leftx, 2)
    right_fit = np.polyfit(righty, rightx, 2)
    
    # Sanity Check
    ploty = np.linspace(0, binary_warped.shape[0]-1, binary_warped.shape[0] )
    y_eval = np.max(ploty)
    
    ym_per_pix = 30/720 # meters per pixel in y dimension
    xm_per_pix = 3.7/700 # meters per pixel in x dimension

    # Fit new polynomials to x,y in world space
    left_fit_cr = np.polyfit(lefty*ym_per_pix, leftx*xm_per_pix, 2)
    right_fit_cr = np.polyfit(righty*ym_per_pix, rightx*xm_per_pix, 2)
    # Calculate the new radii of curvature
    left_curverad = ((1 + (2*left_fit_cr[0]*y_eval*ym_per_pix + left_fit_cr[1])**2)**1.5) / np.absolute(2*left_fit_cr[0])
    right_curverad = ((1 + (2*right_fit_cr[0]*y_eval*ym_per_pix + right_fit_cr[1])**2)**1.5) / np.absolute(2*right_fit_cr[0])
    # Now our radius of curvature is in meters
    avg_radius = (left_curverad + right_curverad) / 2
# #     print(left_curverad, 'm', right_curverad, 'm', avg_radius, 'm')
#     if abs(left_curverad-right_curverad)>250 and abs(left_curverad-right_curverad)<300:
#         if left_curverad < right_curverad:
#             right_fit[:2] = left_fit[:2]
# #             right_fit[2] = left_fit[2] + 6.8 * 1e2
#         else:
#             left_fit[:2] = right_fit[:2] 
# #             left_fit[2] = right_fit[2] - 6.8 * 1e2
    
    left_fitx = left_fit[0]*y_eval**2 + left_fit[1]*y_eval + left_fit[2]
    right_fitx = right_fit[0]*y_eval**2 + right_fit[1]*y_eval + right_fit[2]

    left_line_pos = left_fitx*xm_per_pix
    right_line_pos = right_fitx*xm_per_pix
    lane_center = (right_line_pos-left_line_pos)/2
    center_offset = (binary_warped.shape[0]/2*xm_per_pix)-lane_center
    
#     skip = False
#     if abs(left_curverad-right_curverad)>250 and abs(left_curverad-right_curverad)<300:
#         skip = True
        
    y_eval = y_eval//2
    left_fitx = left_fit[0]*y_eval**2 + left_fit[1]*y_eval + left_fit[2]
    right_fitx = right_fit[0]*y_eval**2 + right_fit[1]*y_eval + right_fit[2]
    left_line_pos = left_fitx*xm_per_pix
    right_line_pos = right_fitx*xm_per_pix
    lane_width = abs(right_line_pos-left_line_pos)
    
# #     print(lane_width)
#     if not 3.25< lane_width < 3.35:
#         skip = True
    return left_fit, right_fit, out_img, left_lane_inds, right_lane_inds, \
            avg_radius, center_offset, lane_width #skip

Results Visualization

In [294]:
# Generate x and y values for plotting
src_img = cv2.imread('test_images/test2.jpg')
src_img = cv2.undistort(src_img, mtx, dist, None, mtx)
src_img = cv2.cvtColor(src_img, cv2.COLOR_BGR2RGB)
color_binary, combined_binary = color_grad_thr(src_img, ksize=7, s_thresh=(190, 255), 
                                               sx_thresh=(30, 150))
src_img_wraped = warper(src_img, src_pts, dst_pts)
binary_warped = warper(combined_binary, src_pts, dst_pts)

left_fit, right_fit, out_img, left_lane_inds, right_lane_inds, \
    _, _, _ = lane_detection(binary_warped)

nonzero = binary_warped.nonzero()
nonzeroy = np.array(nonzero[0])
nonzerox = np.array(nonzero[1])
    
ploty = np.linspace(0, binary_warped.shape[0]-1, binary_warped.shape[0] )
left_fitx = left_fit[0]*ploty**2 + left_fit[1]*ploty + left_fit[2]
right_fitx = right_fit[0]*ploty**2 + right_fit[1]*ploty + right_fit[2]

# print(left_fit)
# print(right_fit)

out_img[nonzeroy[left_lane_inds], nonzerox[left_lane_inds]] = [255, 0, 0]
out_img[nonzeroy[right_lane_inds], nonzerox[right_lane_inds]] = [0, 0, 255]
plt.imshow(out_img)
plt.plot(left_fitx, ploty, color='yellow')
plt.plot(right_fitx, ploty, color='yellow')
# plt.plot(binary_warped.shape[0]-histogram, color='cyan', linewidth=2)
plt.xlim(0, 1280)
plt.ylim(720, 0)
Out[294]:
(720, 0)

4.2 Skipping the sliding windows step once we know where the lines are

In [8]:
# Assume you now have a new warped binary image 
# from the next frame of video (also called "binary_warped")
# It's now much easier to find line pixels!
def skip_windows_step(binary_warped, left_fit, right_fit, margin = 100):
    nonzero = binary_warped.nonzero()
    nonzeroy = np.array(nonzero[0])
    nonzerox = np.array(nonzero[1])

    left_lane_inds = ((nonzerox > (left_fit[0]*(nonzeroy**2) + left_fit[1]*nonzeroy + left_fit[2] - margin)) & (nonzerox < (left_fit[0]*(nonzeroy**2) + left_fit[1]*nonzeroy + left_fit[2] + margin))) 
    right_lane_inds = ((nonzerox > (right_fit[0]*(nonzeroy**2) + right_fit[1]*nonzeroy + right_fit[2] - margin)) & (nonzerox < (right_fit[0]*(nonzeroy**2) + right_fit[1]*nonzeroy + right_fit[2] + margin)))  

    # Again, extract left and right line pixel positions
    leftx = nonzerox[left_lane_inds]
    lefty = nonzeroy[left_lane_inds] 
    rightx = nonzerox[right_lane_inds]
    righty = nonzeroy[right_lane_inds]
    # Fit a second order polynomial to each
    new_left_fit = np.polyfit(lefty, leftx, 2)
    new_right_fit = np.polyfit(righty, rightx, 2)
    
    return new_left_fit, new_right_fit

Results Visualization

In [295]:
margin = 100
new_left_fit, new_right_fit = skip_windows_step(binary_warped, left_fit, 
                                                right_fit, margin=margin)

# Generate x and y values for plotting
ploty = np.linspace(0, binary_warped.shape[0]-1, binary_warped.shape[0] )
left_fitx = left_fit[0]*ploty**2 + left_fit[1]*ploty + left_fit[2]
right_fitx = right_fit[0]*ploty**2 + right_fit[1]*ploty + right_fit[2]

# Create an image to draw on and an image to show the selection window
out_img = np.dstack((binary_warped, binary_warped, binary_warped))*255
window_img = np.zeros_like(out_img)
# Color in left and right line pixels
out_img[nonzeroy[left_lane_inds], nonzerox[left_lane_inds]] = [255, 0, 0]
out_img[nonzeroy[right_lane_inds], nonzerox[right_lane_inds]] = [0, 0, 255]

# Generate a polygon to illustrate the search window area
# And recast the x and y points into usable format for cv2.fillPoly()
left_line_window1 = np.array([np.transpose(np.vstack([left_fitx-margin, ploty]))])
left_line_window2 = np.array([np.flipud(np.transpose(np.vstack([left_fitx+margin, ploty])))])
left_line_pts = np.hstack((left_line_window1, left_line_window2))
right_line_window1 = np.array([np.transpose(np.vstack([right_fitx-margin, ploty]))])
right_line_window2 = np.array([np.flipud(np.transpose(np.vstack([right_fitx+margin, ploty])))])
right_line_pts = np.hstack((right_line_window1, right_line_window2))

# Draw the lane onto the warped blank image
cv2.fillPoly(window_img, np.int_([left_line_pts]), (0,255, 0))
cv2.fillPoly(window_img, np.int_([right_line_pts]), (0,255, 0))
result = cv2.addWeighted(out_img, 1, window_img, 0.3, 0)
plt.imshow(result)
plt.plot(left_fitx, ploty, color='yellow')
plt.plot(right_fitx, ploty, color='yellow')
plt.xlim(0, 1280)
plt.ylim(720, 0)

Out[295]:
(720, 0)

4.3 Testing the Convoultion Method

In [9]:
def window_mask(width, height, img_ref, center,level):
    output = np.zeros_like(img_ref)
    output[int(img_ref.shape[0]-(level+1)*height):int(img_ref.shape[0]-level*height),max(0,int(center-width/2)):min(int(center+width/2),img_ref.shape[1])] = 1
    return output

def find_window_centroids(warped, window_width, window_height, margin):
    window_centroids = [] # Store the (left,right) window centroid positions per level
    window = np.ones(window_width) # Create our window template that we will use for convolutions
    
    # First find the two starting positions for the left and right lane by using np.sum to get the vertical image slice
    # and then np.convolve the vertical image slice with the window template 
    
    # Sum quarter bottom of image to get slice, could use a different ratio
    l_sum = np.sum(warped[int(3*warped.shape[0]/4):,:int(warped.shape[1]/2)], axis=0)
    l_center = np.argmax(np.convolve(window,l_sum))-window_width/2
    r_sum = np.sum(warped[int(3*warped.shape[0]/4):,int(warped.shape[1]/2):], axis=0)
    r_center = np.argmax(np.convolve(window,r_sum))-window_width/2+int(warped.shape[1]/2)
    
    # Add what we found for the first layer
    window_centroids.append((l_center,r_center))
    
    # Go through each layer looking for max pixel locations
    for level in range(1,(int)(warped.shape[0]/window_height)):
	    # convolve the window into the vertical slice of the image
	    image_layer = np.sum(warped[int(warped.shape[0]-(level+1)*window_height):int(warped.shape[0]-level*window_height),:], axis=0)
	    conv_signal = np.convolve(window, image_layer)
	    # Find the best left centroid by using past left center as a reference
	    # Use window_width/2 as offset because convolution signal reference is at right side of window, not center of window
	    offset = window_width/2
	    l_min_index = int(max(l_center+offset-margin,0))
	    l_max_index = int(min(l_center+offset+margin,warped.shape[1]))
	    l_center = np.argmax(conv_signal[l_min_index:l_max_index])+l_min_index-offset
	    # Find the best right centroid by using past right center as a reference
	    r_min_index = int(max(r_center+offset-margin,0))
	    r_max_index = int(min(r_center+offset+margin,warped.shape[1]))
	    r_center = np.argmax(conv_signal[r_min_index:r_max_index])+r_min_index-offset
	    # Add what we found for that layer
	    window_centroids.append((l_center,r_center))

    return window_centroids

def lane_detection_conv(warped):
    # window settings
    window_width = 50 
    window_height = 90 #72 #80 # Break image into 9 vertical layers since image height is 720
    margin = 100 # How much to slide left and right for searching

    window_centroids = find_window_centroids(warped, window_width, window_height, margin)

    # If we found any window centers
    if len(window_centroids) > 0:

        # Points used to draw all the left and right windows
        l_points = np.zeros_like(warped)
        r_points = np.zeros_like(warped)

        # Go through each level and draw the windows 	
        for level in range(0,len(window_centroids)):
            # Window_mask is a function to draw window areas
            l_mask = window_mask(window_width,window_height,warped,window_centroids[level][0],level)
            r_mask = window_mask(window_width,window_height,warped,window_centroids[level][1],level)
            # Add graphic points from window mask here to total pixels found 
            l_points[(l_points == 255) | ((l_mask == 1) ) ] = 255
            r_points[(r_points == 255) | ((r_mask == 1) ) ] = 255

        # Draw the results
        template = np.array(r_points+l_points,np.uint8) # add both left and right window pixels together
        zero_channel = np.zeros_like(template) # create a zero color channle 
        template = np.array(cv2.merge((zero_channel,template,zero_channel)),np.uint8) # make window pixels green
        warpage = np.array(cv2.merge((warped,warped,warped)),np.uint8) # making the original road pixels 3 color channels
        output = cv2.addWeighted(warpage, 1, template, 0.5, 0.0) # overlay the orignal road image with window results

    # If no window centers found, just display orginal road image
    else:
        output = np.array(cv2.merge((warped,warped,warped)),np.uint8)
        
    return output

Results Visualization

In [297]:
output = lane_detection_conv(binary_warped)
# Display the final results
plt.imshow(output)
plt.title('window fitting results')
plt.show()

5. Calculate Radius of Curvature

5.1 Pixel Space

In [26]:
# Define y-value where we want radius of curvature
# I'll choose the maximum y-value, corresponding to the bottom of the image
y_eval = np.max(ploty)
left_curverad = ((1 + (2*left_fit[0]*y_eval + left_fit[1])**2)**1.5) / np.absolute(2*left_fit[0])
right_curverad = ((1 + (2*right_fit[0]*y_eval + right_fit[1])**2)**1.5) / np.absolute(2*right_fit[0])
print(left_curverad, right_curverad)
# Example values: 1926.74 1908.48
1365.30524175 2263.21317422

5.2 Real World Space

In [35]:
# Define conversions in x and y from pixels space to meters
ym_per_pix = 30/720 # meters per pixel in y dimension
xm_per_pix = 3.7/700 # meters per pixel in x dimension

# Fit new polynomials to x,y in world space
left_fit_cr = np.polyfit(lefty*ym_per_pix, leftx*xm_per_pix, 2)
right_fit_cr = np.polyfit(righty*ym_per_pix, rightx*xm_per_pix, 2)
# Calculate the new radii of curvature
left_curverad = ((1 + (2*left_fit_cr[0]*y_eval*ym_per_pix + left_fit_cr[1])**2)**1.5) / np.absolute(2*left_fit_cr[0])
right_curverad = ((1 + (2*right_fit_cr[0]*y_eval*ym_per_pix + right_fit_cr[1])**2)**1.5) / np.absolute(2*right_fit_cr[0])
# Now our radius of curvature is in meters
avg_radius = (left_curverad + right_curverad) / 2
print(left_curverad, 'm', right_curverad, 'm', avg_radius, 'm')
# Example values: 632.1 m    626.2 m
430.566783423 m 694.822197669 m 562.694490546 m

6. Output visual display of the lane boundaries and numerical estimation of lane curvature and vehicle position.

In [10]:
# Create an image to draw the lines on
warp_zero = np.zeros_like(binary_warped).astype(np.uint8)
color_warp = np.dstack((warp_zero, warp_zero, warp_zero))

# Recast the x and y points into usable format for cv2.fillPoly()
pts_left = np.array([np.transpose(np.vstack([left_fitx, ploty]))])
pts_right = np.array([np.flipud(np.transpose(np.vstack([right_fitx, ploty])))])
pts = np.hstack((pts_left, pts_right))

# Draw the lane onto the warped blank image
cv2.fillPoly(color_warp, np.int_([pts]), (0,255, 0))

# Warp the blank back to original image space using inverse perspective matrix (Minv)
Minv = cv2.getPerspectiveTransform(dst_pts, src_pts)
newwarp = cv2.warpPerspective(color_warp, Minv, (binary_warped.shape[1], binary_warped.shape[0])) 
# Combine the result with the original image
result = cv2.addWeighted(src_img, 1, newwarp, 0.3, 0)

left_fit, right_fit, _, _, _, avg_radius, center_offset, _ = lane_detection(binary_warped)

font = cv2.FONT_HERSHEY_SIMPLEX
if avg_radius > 3000:
    cv2.putText(result,'Lane nearly straight',(30,80), font, 2,(255,255,255),3,cv2.LINE_AA)
else:
    cv2.putText(result,'Radius of Curvature = {:0.2f}m'.format(avg_radius),(30,80), font, 2,(255,255,255),3,cv2.LINE_AA)

if center_offset >= 0:
    cv2.putText(result,'Vehicle is {:0.2f}m left of center'.format(center_offset),(30,160), font, 2,(255,255,255),3,cv2.LINE_AA)
else:
    cv2.putText(result,'Vehicle is {:0.2f}m right of center'.format(abs(center_offset)),(30,160), font, 2,(255,255,255),3,cv2.LINE_AA)
 
plt.imshow(result)
---------------------------------------------------------------------------
NameError                                 Traceback (most recent call last)
<ipython-input-10-d90fe8d7b427> in <module>()
      1 # Create an image to draw the lines on
----> 2 warp_zero = np.zeros_like(binary_warped).astype(np.uint8)
      3 color_warp = np.dstack((warp_zero, warp_zero, warp_zero))
      4 
      5 # Recast the x and y points into usable format for cv2.fillPoly()

NameError: name 'binary_warped' is not defined

7. Test Pipeline On a Video

In [21]:
# Import everything needed to edit/save/watch video clips
from moviepy.editor import VideoFileClip
from IPython.display import HTML
from collections import deque
from itertools import accumulate

def process_image(image, my_buffer=deque()):
    buffer_len = 5
    curr_buffer_len = len(my_buffer)
    # Just start with any of lane detection methods (wheather plain or using the convolution method)
    # and don't try the skip method now
    _, combined_binary = color_grad_thr(image, s_thresh=(190, 255), #s_thresh=(205, 255), #s_thresh=(190, 255),
                                        sx_thresh=(50, 150))#, ksize=3, sx_thresh=(30, 150))
    binary_warped = warper(combined_binary, src_pts, dst_pts)

    # Create an image to draw the lines on
    warp_zero = np.zeros_like(binary_warped).astype(np.uint8)
    color_warp = np.dstack((warp_zero, warp_zero, warp_zero))

    # TODO: clean up lane_detection return values
    # Use delta change in lane width instead of abs value of lane width
    left_fit, right_fit, _, _, _, radius, center_offset, lane_width = lane_detection(binary_warped)
    
    avg_lane_width = lane_width
    avg_left_fit, avg_right_fit, avg_radius = left_fit, right_fit, radius
    
    if curr_buffer_len > buffer_len - 1:
        my_buffer.popleft()
        curr_buffer_len = curr_buffer_len - 1
        
    if curr_buffer_len > 0:
        my_buffer_avg = [item/curr_buffer_len for item in list(zip(*map(accumulate, zip(*my_buffer))))[-1]]
        avg_left_fit, avg_right_fit, avg_radius, avg_lane_width = my_buffer_avg
#         _, _, _, avg_lane_width = my_buffer_avg
#         avg_left_fit, avg_right_fit, avg_radius, avg_lane_width = my_buffer[-1]


    delta_lane_width = abs(avg_lane_width - lane_width)
    if delta_lane_width < 0.2:
        my_buffer.append([left_fit, right_fit, radius, lane_width])
        curr_buffer_len = curr_buffer_len + 1
        my_buffer_avg = [item/curr_buffer_len for item in list(zip(*map(accumulate, zip(*my_buffer))))[-1]]
        avg_left_fit, avg_right_fit, avg_radius, avg_lane_width = my_buffer_avg

    left_fit, right_fit, radius = avg_left_fit, avg_right_fit, avg_radius
    
    ploty = np.linspace(0, binary_warped.shape[0]-1, binary_warped.shape[0] )
    left_fitx = left_fit[0]*ploty**2 + left_fit[1]*ploty + left_fit[2]
    right_fitx = right_fit[0]*ploty**2 + right_fit[1]*ploty + right_fit[2]

    # Recast the x and y points into usable format for cv2.fillPoly()
    pts_left = np.array([np.transpose(np.vstack([left_fitx, ploty]))])
    pts_right = np.array([np.flipud(np.transpose(np.vstack([right_fitx, ploty])))])
    pts = np.hstack((pts_left, pts_right))

    # Draw the lane onto the warped blank image
    cv2.fillPoly(color_warp, np.int_([pts]), (0,255, 0))

    # Warp the blank back to original image space using inverse perspective matrix (Minv)
    Minv = cv2.getPerspectiveTransform(dst_pts, src_pts)
    newwarp = cv2.warpPerspective(color_warp, Minv, (image.shape[1], image.shape[0])) 
    # Combine the result with the original image
    result = cv2.addWeighted(image, 1, newwarp, 0.3, 0)

    font = cv2.FONT_HERSHEY_SIMPLEX
    if radius > 3000:
        cv2.putText(result,'Lane nearly straight',(30,80), font, 2,(255,255,255),3,cv2.LINE_AA)
    else:
        cv2.putText(result,'Radius of Curvature = {:0.2f}m'.format(radius),(30,80), font, 2,(255,255,255),3,cv2.LINE_AA)
        
    if center_offset >= 0:
        cv2.putText(result,'Vehicle is {:0.2f}m right of center'.format(center_offset),(30,160), font, 2,(255,255,255),3,cv2.LINE_AA)
    else:
        cv2.putText(result,'Vehicle is {:0.2f}m left of center'.format(center_offset),(30,160), font, 2,(255,255,255),3,cv2.LINE_AA)
        
    cv2.putText(result,'Delta Lane Width is {:0.2f}m '.format(delta_lane_width),(30,240), font, 2,(255,255,255),3,cv2.LINE_AA)
#     cv2.putText(result,'left_fitx is: ' + str(left_fit),(30,320), font, 1,(255,255,255),3,cv2.LINE_AA)

    return result

Let's try the project_video.mp4 first ...

In [22]:
project_output = 'project_output3.mp4'
clip1 = VideoFileClip("project_video.mp4")

# project_output = 'challenge_output.mp4'
# clip1 = VideoFileClip("challenge_video.mp4")

# project_output = 'harder_challenge_output.mp4'
# clip1 = VideoFileClip("harder_challenge_video.mp4")


# project_output = 'test_output.mp4'
# clip1 = VideoFileClip("test_video.mp4")

white_clip = clip1.fl_image(process_image) #NOTE: this function expects color images!!
%time white_clip.write_videofile(project_output, audio=False, threads=8)
[MoviePy] >>>> Building video project_output3.mp4
[MoviePy] Writing video project_output3.mp4
100%|█████████▉| 1260/1261 [05:39<00:00,  3.43it/s]
[MoviePy] Done.
[MoviePy] >>>> Video ready: project_output3.mp4 

CPU times: user 16min 17s, sys: 2min 13s, total: 18min 31s
Wall time: 5min 40s
In [23]:
HTML("""
<video width="960" height="540" controls>
  <source src="{0}">
</video>
""".format(project_output))
Out[23]:
In [ ]: